EFDA–JET–PR(12)41 A.E. Shevelev, I.N. Chugunov, V.G. Kiptily, D.N. Doinikov, D.B. Gin, E.M. Khilkevitch, V.O. Naidenov and JET EFDA contributors Reconstruction of Distribution Functions of Fast Ions and Runaway Electrons in ITER Plasmas Using Gamma-Ray Spectrometry “This document is intended for publication in the open literature. It is made available on the understanding that it may not be further circulated and extracts or references may not be published prior to publication of the original when applicable, or without the consent of the Publications Officer, EFDA, Culham Science Centre, Abingdon, Oxon, OX14 3DB, UK.” “Enquiries about Copyright and reproduction should be addressed to the Publications Officer, EFDA, Culham Science Centre, Abingdon, Oxon, OX14 3DB, UK.” The contents of this preprint and all other JET EFDA Preprints and Conference Papers are available to view online free at www.iop.org/Jet. This site has full search facilities and e-mail alert options. The diagrams contained within the PDFs on this site are hyperlinked from the year 1996 onwards. Reconstruction of Distribution Functions of Fast Ions and Runaway Electrons in ITER Plasmas Using Gamma-Ray Spectrometry A.E. Shevelev1, I.N. Chugunov1, V.G. Kiptily2, D.N. Doinikov1, D.B. Gin1, E.M. Khilkevitch1, V.O. Naidenov1 and JET EFDA contributors* JET-EFDA, Culham Science Centre, OX14 3DB, Abingdon, UK 1Ioffe Physical-Technical Institute of the Russian Academy of Sciences, Polytechnicheskaya 26, St. Petersburg, 194021, Russia 2EURATOM-CCFE Fusion Association, Culham Science Centre, OX14 3DB, Abingdon, OXON, UK * See annex of F. Romanelli et al, “Overview of JET Results”, (24th IAEA Fusion Energy Conference, San Diego, USA (2012)). Preprint of Paper to be submitted for publication in Nuclear Fusion . ABSTRACT. Use of gamma-ray spectrometers on ITER could allow solving one of the most important issues for the safe tokamak operations - diagnosing runaway electrons. 2-D hard X-ray (HXR) emission measurements can provide important information on the runaway beam location in the ITER plasmas and allow estimating the value of the runaway current in the MeV energy range. The DEGAS code has been developed for deconvolution of gamma-ray spectra emitted from plasmas. Using the recorded HXR spectra, the code can reconstruct the runaway electron energy distribution. Results of Monte- Carlo modelling of the gamma-ray spectrometer response functions and bremsstrahlung spectra calculated for electrons in wide energy range are used in the DEGAS code. The deconvolution of spectra allows identifying nuclear reactions, which take place during plasma discharges, calculate their gamma-ray line intensities and determine the maximal energy of runaway electrons with accuracy, which satisfies the ITER Project Requirements. The DEGAS code was used for processing of spectra recorded in JET experiments. Application of the deconvolution technique for 2-D gamma- ray emission measurements, which can facilitate reconstruction of fast ion spatial distributions in ITER plasmas, is discussed. 1. INTRODUCTION A Vertical Gamma-Ray Camera (VGC) for ITER is being designed in Ioffe Institute [1]. The gamma- ray spectrometry could provide time and spatially-resolved measurements of the gamma-ray source strength delivering unique information on confined alphas, fast ions and runaway electrons in ITER. Principles of the γ diagnostics are described elsewhere [2-4]. They are based on spectrometry of the gamma radiation generated during nuclear reactions, occurring between fast particles, fusion products, fuel components and plasma impurities. Monitoring of runaway electron generation is one of the most important issues for safe tokamak operations. The diagnosis of runaway electrons in ITER is required for machine protection. The runaway’s energy should be measured in the range up to100MeV with 20% accuracy and 10ms time resolution. The runaway current after thermal quench is controlled with 30% accuracy, and after failed breakdown - in the range up to 1 МА with accuracy 50kA [5]. Diagnosing the energy and spatial distributions of runaway electrons can be carried out with 2-D bremsstrahlung spectrum measurements. Since the gamma diagnostics provide indirect data about explored quantities, such as, parameters of fast ions and runaway electron distributions, correctness of the interpretation of measured gamma and Hard X-ray (HXR) spectra is an important issue to be answered. This paper presents new techniques for experimental data analysis in the framework of ITER gamma-ray diagnostic system design. 2. RECONSTRUCTIONS OF FAST ION DISTRIBUTIONS FROM MEASURED GAMMA RADIATION Reconstruction of an initial radiation spectrum is one of the primary goals of nuclear spectroscopy. 1 Its solution is complicated by the fact that the response of the detector to monochromatic gamma- rays represents a complex function depending on many detector characteristics and experimental conditions. Gamma-ray spectrometric measurements in experiments with hot plasma are especially complicated. Usually gamma-ray diagnosing is carried out at high neutron and continuous gamma- ray background. Therefore, the separation of gamma lines in the recorded spectrum, identification of their energies and finding their intensities is an essential task for the experimental data processing [3]. Attempts of reconstruction of initial energy distribution from the spectra measured by scintillation and semiconductor detectors were carried out earlier, for example, in the work [6]. The problem urgency has led to various techniques of deconvolution of spectra and images. In a number of works a comparison has been made between various methods on purpose to choose the most suitable one for spectra processing. It was shown that maximum likelihood estimation using expectation maximization (ML-EM) method [7], known just as Richardson-Lusi [8,9] method, is one of the best at gamma-ray spectra reconstruction: it shows good stability to the initial data noise and allows rather precise recovery of the initial spectra [6]. In all mentioned works the deconvolution methods were applied to the case when the initial spectrum is discrete. Besides, high statistics of spectra were guessed. At the same time, spectra recorded in fusion plasma experiments on tokamaks sometimes have rather low statistics. Moreover, a contribution of neutron induced radiation is present in the spectra. All these mentioned features make an application of the algorithms in experimental data processing more complicated. ML-EM method has been studied for reconstruction of gamma spectra recorded in hot plasma experiments. Gamma-ray spectrum y (ε) measured by a detector can be represented in the following form +∞ y (ε) = x (ε′) h (ε, ε′) dε′ + n (ε), (1) 0 where x is an initial gamma spectrum, h - detector’s instrumental function, n – noise and e - gamma- ray energy. Deconvolution problem consists of obtaining the initial spectrum x from the measured spectrum y using known instrument function h. Expression (1) can be presented in the matrix form as: y = Hx + n. Iterative algorithm ML-EM for the solution of this problem can be written, as: p yj p –1 . xi = xi Σj hj,i p–1 (2) Σk hj,k xk p p Transformation xi = max (xi , 0) is carried out at every step of data processing. The basic procedure has been modified. For inhibition of oscillations in the presence of a background pedestal it was offered to carry out smoothing procedure at every j iteration that allows reducing the oscillation amplitude. For magnification of the algorithm resolving ability it was proposed to reduce a channel width in the initial spectrum by interpolation. The DEGAS code 2 (DEconvolution of GAmma Spectrum) realizing the deconvolution procedure described above and defining peak intensities in the reconstructed gamma-ray spectra has been developed in Ioffe Institute [10]. To realize the described algorithm it is necessary to carry out calculations of detector instrument functions with realistic geometrical and technical parameters in a wide energy range with as small energy steps as possible. Examples of scintillation detector instrument functions calculated using MCNP (Monte Carlo N-Particle) code are represented in figure 1. In order to examine capabilities of the developed code, some experiments with gamma ray sources were carried out. NaI(Tl) detector ø150x100 mm of size and with energy resolution of 11.5% on 661.6keV line was used in the measurements. Response functions of this detector were calculated by MCNP code in 60–3000keV energy range with 20keV step. In the developed calculation model an isotropic gamma-ray source was placed in 25.5cm apart from the detector. 60Co and 137Cs radioactive sources with known activities of decay were installed the same distance from the real detector for measurements. Result of the measured spectrum processing is shown in figure 2. Accuracy of the sources activities determination was 1.5%. The developed code was successfully used for processing of gamma-ray spectra measured in experiments on JET. Gamma-ray spectrometry system on JET consists of the BGO detector having quasi-tangential line of sight and NaI(Tl) and LaBr3(Ce) spectrometers with vertical line of plasma view. Response functions for BGO detector with a quasi-tangential line of plasma view have been calculated, as well as for other JET spectrometers. A polyethylene attenuator installed in front of the detector was also taken into account when designing the MCNP model in the calculations of the response functions. In figure